Entering edit mode
4.7 years ago
kinalimeric
▴
40
Hi all,
I want to visualize only the top ten peaks using deep tools. However, the resolution is really low if I run computeMatrix with the top 10 peaks even though I increase dpi and changing the —heatmapHeight of the heat-map. Also, I tried to run computeMatrix with all the increasing peaks then get the top 10 from it by arranging —zMin and —zMax but this time I could not understand which peaks I am getting. What should I do to plot heatmap of the top 10 increasing peaks?
Best,
What does "top 10" refer to?
I have 8 samples. I performed differential binding analysis using diffbind. I want to show the peaks that change the most (top10) between those samples.
Using a heatmap? So a full heatmap and then kind of highlight the 10 regions or just the 10 regions alone? When you plot heatmaps you typically transform the log2 normalized counts to the Z-scale. Reason is that read counts can be quite different between regions so that a continuous color scale could be dominated by very high values of a single sample/peak. Not sure deeptools is a good choice here. Without knowing details what you really want to visualize why not just extracting the normalized counts for these regions from your DiffBind object, transform to log2, then to Z-score
t(scale(t(counts)))
and then plot it with any heatmap tool such asComplexHeatmap
.Thank you for your help and explanation.